Addendum to SEDAR16-DW-22

Addendum to SEDAR16-DW-22 Introduction Six king mackerel indices of abundance, two for each region Gulf of Mexico, South Atlantic, and Mixing Zone, were constructed for the SEDAR16 data workshop using coastal logbook commercial fishery data. The regulatory history of the king mackerel commercial fishery was not available until the data workshop began, therefore, possible effects of regulatory measures were not considered in the construction of those initial indices. In addition, reporting king mackerel landings to the coastal logbook program was not required prior to 1998. The effect of that underreporting on the initial indices was unknown. The indices working group recommended several revisions to the commercial logbook indices. These included: Account for regulatory measures in the construction of commercial indices: Ignore minimum size regulations this will be accounted for in the population model Exclude data from closed seasons Examine effects of trip limits on fishing effort and determine the feasibility of index construction Treat individual vessels as repeated measures in the analyses to address possible differences in catch per unit effort (CPUE) among vessels that had reported landings and effort throughout the time series and vessels that provided landings and effort only after reporting became mandatory in 1998. Revised indices, incorporating the SEDAR 16 data workshop recommendations were constructed using the available coastal logbook CPUE series, from 1993-2006. Separate indices were developed for the Gulf of Mexico, South Atlantic, and the king mackerel Mixing Zone. Methods Effects of regulations Examination of the effects of regulatory trip limits on index construction is described in SEDAR16-AW-02. As part of that analysis, trips were categorized by the percentage of king mackerel (by weight) in the trip landings. That categorization was used as a proxy for targeting and was used as a factor in the revised indices. Based upon examination of the regulatory effects, only the 25 fish per trip limit in the Mixing Zone was assumed to effect index construction. Landings and effort data reported from the Mixing Zone during periods with a 25 fish per trip limit were, therefore, excluded from the analysis. In addition, data from periods when the king mackerel fishery was closed were also excluded. No other data restrictions were made due to regulatory measures. Available data For each fishing trip, the coastal logbook database includes a unique trip identifier, the landing date, fishing gear deployed, areas fished (Figure 1), number of days at sea, number of crew, gear specific fishing effort (for hook and line fisheries: number of lines fished, number of hooks per line and estimated total fishing time), species caught and whole weight of the landings. Multiple areas fished and multiple gears fished may be recorded for a single fishing trip. In such cases, assigning catch and effort to specific locations or gears was not possible; therefore, only trips which reported one area and one gear fished were included in these analyses. Only data from hook and line fisheries were used in these analyses. Hook and line catch rate was calculated in weight of fish per hook-hour. For each trip, CPUE was calculated as: 1

CPUE = total pounds of king mackerel/(number of lines fished*number of hooks per line*total hours fished) Three regions were defined (Figure 1) in the analyses. The Gulf of Mexico included all areas from southwest Florida to Mexico other than areas 1-3. The south Atlantic was defined as the area north of 30 o N to 37 o N. The Mixing Zone was defined as the area south of 30 o N to 24 o N in the south Atlantic and including Gulf of Mexico fishing areas 1-3. Data used in constructing the commercial hook and line fishery indices of abundance were limited to catch and effort reported from vessels that together accounted for the highest 80% of the reported hook and line gear landings of king mackerel over the period 1993-2006. The selection of vessels was made for each region by ordering all vessels firstly by the number of years each reported king mackerel landings in the region and secondly by the vessel s total king mackerel landings from the region. For example, vessels that reported king mackerel landings in 14 years during 1993-2006 in the Mixing Zone were ordered by their total reported king mackerel landings in the Mixing Zone followed by vessels that reported king mackerel landings in 13 years. Vessels were added to a region-specific data set until 80% of the total king mackerel landings from a region were accounted for by the landings reported by those included vessels. Once the vessel list for each region was defined, all hook and line gear trips within each region reported by the selected vessels were considered potential king mackerel trips and were included in the analyses. Clear outliers in the data, i.e. values falling outside the 99.5 percentile of the data, were excluded from the analyses. These included data from trips reporting more than seven lines fished, 20 hooks per line fished, or more than 10 days at sea. Index Development Ten factors were considered as possible influences on both the proportion of trips that landed king mackerel and the catch rate of king mackerel. In order to develop a well -balanced sample design, the ten factors were defined as: Gulf of Mexico Factor Levels Values Year 14 1993-2006 Area 9 Gulf of Mexico shrimp grids 4-5, 6-7, 8, 9, 10-12, 13, 14-15, 16-17, 18-21 see Figure 1. Days at sea (AWAY1)* 4 1, 2, 3, 4-10 Target 4 <25% of catch was king mackerel, 25-50%, 50-75%, >75% Crew (CREW1) 4 1, 2, 3, or 4+ crew members Vessel length (VES_LEN) 4 35 feet, >35 to 45, >45, unknown Number of lines fished 4 1-2, 3, 4, 5-7 (NUMGEAR1) Number of hooks/line 5 1, 2, 3-10, 11-15, 16-20 (EFFORT1) Gear 2 Handline (includes electric reels), trolling Hours fished (Hrs_fished) 6 6, >6-8, >8-12, >12-24, >24-48, >48 *Names in parentheses appear in some figures and tables. 2

Mixing Zone Factor Levels Value Year 14 1993-2006 Area* 10 Areas1-2 and 2482; 3; 2479-2480; 2481; 2575-2580; 2674-2679; 2680; 2777-2779; 2780-2781; 2842-2981 see Figure 1. Days at sea (AWAY1)** 2 1, 2-10 Target 4 <25% of catch was king mackerel, 25-50%, 50-75%, >75% Crew (CREW1) 2 1, 2+ crew members Vessel length (VES_LEN) 5 25 feet, >25-30, >30 to 35, >35, unknown Number of lines fished (NUMGEAR1) 4 1, 2, 3, 4-7 Number of hooks/line (EFFORT1) 2 1, 2-20 Gear 2 Handline (includes electric reels), trolling Hours fished (Hrs_fished) 5 5, >5-7, >7-8, >8-10, >10 *Areas 1-2 and 2482 were combined. **Names in parentheses appear in some figures and tables. South Atlantic Factor Levels Value Year 14 1993-2006 Area 5 Areas 3075-3280; 3370-3379; 3470-3476; 3477-3478; 3570-3677 see Figure 1. Days at sea (AWAY1)* 3 1, 2-3, 4-10 Target 4 <25% of catch was king mackerel, 25-50%, 50-75%, >75% Crew (CREW1) 3 1, 2, 3+ crew members Vessel length (VES_LEN) 4 30 feet, >30-35, >35, unknown Number of lines fished 3 1-2, 3, 4-7 (NUMGEAR1) Number of hooks/line 3 1, 2, 3-20 (EFFORT1) Gear 2 Handline (includes electric reels), trolling Hours fished (Hrs_fished) 6 6, >6-8, >8-12, >12-24, >24-48, >48 *Names in parentheses appear in some figures and tables. The delta lognormal model approach (Lo et al. 1992) was used to construct standardized indices of abundance. This method combines separate generalized linear model (GLM) analyses of the proportion of successful trips (trips that landed king mackerel) and the catch rates on successful trips to construct a single standardized CPUE index. Parameterization of each model was accomplished using a GLM procedure (GENMOD; Version 8.02 of the SAS System for Windows 2000. SAS Institute Inc., Cary, NC, USA). For each GLM analysis of proportion positive trips, a type-3 model was fit, a binomial error distribution was assumed, and the logit link was selected. The response variable was proportion successful trips. During the analysis of catch rates on successful trips, a type-3 model assuming lognormal error distribution was examined. The linking function selected was normal, and the response variable was log(cpue). The response variable was calculated as: log(cpue) = ln(pounds of king mackerel/hook-hours). All two-way interactions among significant main effects were examined. A forward stepwise regression procedure was used to determine the set of fixed factors and interaction terms that explained a significant portion of the observed variability. Each potential factor was added to the null 3

model sequentially and the resulting reduction in deviance per degree of freedom was examined. The factor that caused the greatest reduction in deviance per degree of freedom was added to the base model if the factor was significant based upon a Chi-Square test (p<0.05), and the reduction in deviance per degree of freedom was 1%. This model then became the base model, and the process was repeated, adding factors and interactions individually until no factor or interaction met the criteria for incorporation into the final model. Higher order interaction terms were not examined. Once a set of fixed factors was identified, the influence of the YEAR*FACTOR interactions were examined. YEAR*FACTOR interaction terms were included in the model as random effects. Selection of the final mixed model was based on the Akaike s Information Criterion (AIC), Schwarz s Bayesian Criterion (BIC), and a chisquare test of the difference between the 2 log likelihood statistics between successive model formulations (Littell et al. 1996). The final delta-lognormal model was fit using a SAS macro, GLIMMIX (Russ Wolfinger, SAS Institute). All factors were modeled as fixed effects except two-way interaction terms containing YEAR which were modeled as random effects. Individual vessels were included as repeated measures terms. To facilitate visual comparison, a relative index and relative nominal CPUE series were calculated by dividing each value in the series by the mean value of the series. Results and Discussion The final models for the binomial on proportion positive trips and the lognormal on CPUE of successful trips were: Gulf of Mexico 1993-2006: PPT = GEAR + AREA + HOOKS/LINE + YEAR + LINES FISHED + VESSEL LENGTH + AREA*LINES FISHED + AREA*HOOKS/LINE + AREA*VESSEL LENGTH LOG(CPUE) = TARGET + HOOKS/LINE + AREA + HOURS FISHED + LINES FISHED + VESSEL LENGTH + NUMBER OF CREW + YEAR + TARGET*HOURS FISHED + AREA*VESSEL LENGTH + AREA*YEAR + AREA*HOURS FISHED + AREA*LINES FISHED + HOOKS/LINE*AREA The linear regression statistics and analysis of the mixed model formulations of the final models are summarized in Table 1. Mixing Zone 1993-2006: PPT = GEAR + AREA + LINES FISHED + GEAR*AREA + AREA*LINES FISHED LOG(CPUE) = TARGET + LINES FISHED + HOOKS/LINE + AREA + YEAR + HOURS FISHED + VESSEL LENGTH + AREA*YEAR + AREA*HOURS FISHED Year was included in the final binomial portion of the model. The linear regression statistics and analysis of the mixed model formulations of the final GLM models are summarized in Table 2. South Atlantic 1993-2006: PPT = GEAR + LINES FISHED + AREA + YEAR + HOOKS/LINE + AREA*HOOKS/LINE + AREA*YEAR LOG(CPUE) = TARGET + HOOKS/LINE + HOURS FISHED + LINES FISHED + AREA + TARGET*HOURS FISHED + HOOKS/LINE*AREA + LINES FISHED*AREA 4

Year was included in the final lognormal portion of the model. The linear regression statistics and analysis of the mixed model formulations of the final GLM models are summarized in Table 3. Relative nominal CPUE, number of trips, proportion positive trips, and relative abundance indices are provided in Table 4 for Gulf of Mexico king mackerel, Table 5 for the Mixing Zone, and Table 6 for the south Atlantic. The delta-lognormal abundance indices developed for each region and time series, with 95% confidence intervals, are shown in Figures 2-4. In constructing the Gulf of Mexico and Mixing Zone indices, the GLMMIX models failed to converge under the full models described above. With the Gulf of Mexico index, the GLMMIX model failed to converge when the interaction terms were included in either the binomial or lognormal models. Only main effects were included in developing the Gulf of Mexico index. Small sample size, resulting in the inclusion of many factors, likely caused the lack of convergence in the GLMMIX models. Similarly, construction of the Mixing Zone could be completed only when interaction terms were excluded from the final binomial and lognormal models. Plots of the proportion of positive trips per year, nominal CPUE, frequency distributions of the proportion of positive trips, frequency distributions of log(cpue) for positive catch, cumulative normalized residuals, and plots of chi-square residuals by each main effect for the binomial and lognormal models are shown in Figures 5-8 (Gulf of Mexico), Figures 9-12 (Mixing Zone), and Figures 13-16 (South Atlantic). Those diagnostic plots indicate that the fit of the data to the lognormal and binomial models was acceptable. There were some outliers among these data, however, and the frequency distribution of log(cpue) from the Gulf of Mexico and South Atlantic data differed somewhat from the expected normal distribution. Those variations from the expected fit of the data were not sufficient to violate assumptions of the analyses. Standardized catch rates for king mackerel in the Gulf of Mexico had no clear trend over the complete time series (Figure 2). During the period 1993-1996 the mean annual cpue increased slightly. The trend in yearly mean cpue during the period 1998-2006 had decreasing cpues over time. Coefficients of variation were highest during the initial years of the series (Table 4). An overall increase in yearly mean standardized cpue was found for the Mixing Zone (Figure 3). Although there was some variation among years, the highest cpues were found in the last few years of the time series and the lowest cpues occurred during the earliest years of the series. Coefficients of variation were highest during the first three years of the index (Table 5). The index constructed for the south Atlantic indicated no strong trend in yearly mean cpue (Figure 4). Additionally, the nominal cpue series differed from the standardized cpue series. Nominal cpue was lower in the initial years of the series and higher in the final years of the series compared to the standardized cpues. The mean annual nominal cpue showed a clear increase over the 14 year series. The highest annual standardized cpues, however, were found during the first two years of the series. The marked differences in the nominal and standardized series require additional investigation. Coefficients of variation were consistently largest during the final five years of the time series (Table 6). Literature Cited Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D Wolfinger. 1996. SAS System for Mixed Models, Cary NC, USA:SAS Institute Inc., 1996. 663 pp. Lo, N.C., L.D. Jackson, J.L. Squire. 1992. Indices of relative abundance from fish spotter data based on deltalognormal models. Can. J. Fish. Aquat. Sci. 49: 2515-2526. 5

Table 1. Linear regression statistics for the revised GLM models on proportion positive trips (A) and catch rates on positive trips (B) for king mackerel in the Gulf of Mexico data from commercial vessels reporting hook and line gear landings and effort 1993-2006. See text for factor (effect) definitions. A. Effect Num DF Type 3 Tests of Fixed Effects Den DF Chi-Square F Value Pr > ChiSq Pr > F year 13 2045 99.03 7.62 <.0001 <.0001 gear 1 324 604.75 604.75 <.0001 <.0001 area1 8 652 257.02 32.13 <.0001 <.0001 effort1 3 936 186.21 62.07 <.0001 <.0001 numgear1 3 1068 85.37 28.46 <.0001 <.0001 ves_len 3 2045 75.76 25.25 <.0001 <.0001 B. Effect Type 3 Tests of Fixed Effects Num DF Den DF F Value Pr > F year 13 1596 3.02 0.0002 target 3 710 2485.97 <.0001 effort1 3 433 1048.50 <.0001 area1 8 338 72.57 <.0001 hrs_fished 5 1670 110.19 <.0001 numgear1 3 540 182.47 <.0001 ves_len 3 1596 12.88 <.0001 crew1 3 680 18.80 <.0001 6

Table 2. Linear regression statistics for the revised GLM models on proportion positive trips (A) and catch rates on positive trips (B) for king mackerel in the Mixing Zone data from commercial vessels reporting hook and line gear landings and effort 1993-2006. See text for factor (effect) definitions. A. Effect Num DF Type 3 Tests of Fixed Effects Den DF Chi-Square F Value Pr > ChiSq Pr > F year 13 5266 229.39 17.65 <.0001 <.0001 gear 1 1176 2594.16 2594.16 <.0001 <.0001 area1 9 1989 2479.00 275.44 <.0001 <.0001 numgear1 3 4186 533.74 177.91 <.0001 <.0001 B. Effect Type 3 Tests of Fixed Effects Num DF Den DF F Value Pr > F year 13 4445 10.33 <.0001 target 3 5048 13949.1 <.0001 numgear1 3 2054 1234.02 <.0001 effort1 1 946 5173.23 <.0001 area1 9 1164 39.06 <.0001 hrs_fished 4 5894 102.19 <.0001 ves_len 4 4445 31.72 <.0001 7

Table 3. Linear regression statistics for the revised GLM models on proportion positive trips (A) and catch rates on positive trips (B) for king mackerel in the south Atlantic data from commercial vessels reporting hook and line gear landings and effort 1993-2006. (C) The likelihood ratio was used to test the difference of 2 REM log likelihood between two nested models. The final binomial model is indicated with gray shading. See text for factor (effect) definitions. A. Effect Num DF Type 3 Tests of Fixed Effects Den DF Chi-Square F Value Pr > ChiSq Pr > F year 13 52 34.99 2.69 0.0008 0.0057 gear 1 7080 1329.43 1329.43 <.0001 <.0001 numgear1 2 7080 471.17 235.58 <.0001 <.0001 area1 4 52 31.72 7.93 <.0001 <.0001 effort1 2 7080 183.98 91.99 <.0001 <.0001 area1*effort1 8 7080 375.94 46.99 <.0001 <.0001 B. C. Effect ANALYSIS OF MIXED MODEL FORMULATIONS Proportion Positive Type 3 Tests of Fixed Effects Num DF Den DF F Value Pr > F year 13 2393 2.90 0.0003 target 3 1448 2790.30 <.0001 effort1 2 838 981.49 <.0001 hrs_fished 5 2902 11.56 <.0001 numgear1 2 1294 338.93 <.0001 area1 4 451 14.43 <.0001 target*hrs_fished 15 78 51.55 <.0001 effort1*area1 8 2 11.11 0.0852 numgear1*area1 8 12 2.88 0.0485-2 REM Log likelihood Akaike's Information Criterion Schwartz's Bayesian Criterion Likelihood Ratio Test Year+Gear+Numgear1+Area1+Effort1+Area1*Effort1 32029.2 32033.2 32045.0 - - Year+Gear+Numgear1+Area1+Effort1+Area1*Effort1+Area1*Year 31717.2 31723.2 31729.9 312.0 <0.0001 P 8

Table 4. Revised relative nominal CPUE, number of trips, proportion positive trips, and relative abundance index for king mackerel in the Gulf of Mexico. YEAR Relative Nominal CPUE Trips Proportion Successful Trips Relative Index Lower 95% CI (Index) Upper 95% CI (Index) CV (Index) 1993 0.17059 672 0.178571 0.719864 0.553338 0.936506 0.132118 1994 0.36103 552 0.338768 0.881232 0.720546 1.077753 0.100911 1995 0.64503 643 0.399689 0.989605 0.821223 1.192511 0.093458 1996 0.762972 1,131 0.47038 0.973609 0.832576 1.138532 0.078363 1997 1.164979 993 0.594159 1.30679 1.139315 1.498882 0.068654 1998 1.300654 1,464 0.468579 1.288452 1.124224 1.476671 0.068254 1999 1.179421 1,625 0.557538 1.117879 0.98158 1.273105 0.065081 2000 1.577386 1,650 0.638182 1.068021 0.943548 1.208915 0.062017 2001 1.079694 1,713 0.563923 1.055385 0.928151 1.20006 0.064299 2002 1.331662 1,568 0.625638 0.994054 0.879432 1.123617 0.061315 2003 1.364618 1,285 0.654475 0.984621 0.857448 1.130655 0.069231 2004 1.250621 1,260 0.539683 0.923012 0.79699 1.068961 0.073499 2005 0.851199 1,544 0.321244 0.731576 0.607567 0.880896 0.09307 2006 0.960144 1,447 0.418106 0.965899 0.817894 1.140686 0.083307 Table 5. Revised relative nominal CPUE, number of trips, proportion positive trips, and relative abundance index for king mackerel in the Mixing Zone. Year Relative Nominal CPUE Trips Proportion Successful Trips Relative Index Lower 95% CI (Index) Upper 95% CI (Index) CV (Index) 1993 0.403139 4,309 0.247157 0.650916 0.545575 0.776596 0.088442 1994 0.436905 5,162 0.276443 0.658274 0.566274 0.765219 0.075378 1995 0.474768 5,586 0.276942 0.679931 0.586303 0.788512 0.074179 1996 0.623902 6,915 0.345336 0.947111 0.846368 1.059846 0.056276 1997 0.542535 7,350 0.328844 0.805788 0.716925 0.905666 0.058475 1998 1.041776 13,972 0.628614 1.038867 0.952335 1.133263 0.043506 1999 0.998888 15,484 0.570137 1.002828 0.921878 1.090886 0.042102 2000 0.878666 15,145 0.605282 0.930869 0.856611 1.011565 0.041586 2001 0.972751 15,520 0.610438 0.973615 0.897284 1.056438 0.040838 2002 1.050351 14,922 0.603806 1.053141 0.97041 1.142925 0.040924 2003 1.576674 15,224 0.645231 1.277778 1.180265 1.383347 0.039707 2004 1.68558 11,773 0.601376 1.278257 1.171333 1.394941 0.043698 2005 1.450711 10,115 0.573109 1.269829 1.155745 1.395174 0.047095 2006 1.863354 9,934 0.636602 1.432796 1.30433 1.573915 0.046995 9

Table 6. Revised relative nominal CPUE, number of trips, proportion positive trips, and relative abundance index for king mackerel in the south Atlantic. Year Relative Nominal CPUE Trips Proportion Successful Trips Relative Index Lower 95% CI (Index) Upper 95% CI (Index) CV (Index) 1993 0.501149 1,806 0.594684 1.37864 1.183834 1.605503 0.07628 1994 0.444777 2,235 0.501566 1.213336 1.03489 1.422551 0.079665 1995 0.641366 2,815 0.479929 1.122426 0.942012 1.337393 0.087783 1996 0.477344 2,998 0.386258 0.814378 0.651294 1.018299 0.112082 1997 0.858752 3,382 0.466884 1.114656 0.939493 1.322476 0.085637 1998 1.099067 3,998 0.524262 1.022549 0.87639 1.193083 0.077236 1999 1.105474 4,155 0.540554 1.026217 0.876941 1.200903 0.078719 2000 1.074708 4,125 0.578667 1.052164 0.904698 1.223667 0.075609 2001 0.98175 4,256 0.546053 0.909673 0.772893 1.070658 0.081607 2002 0.88771 3,803 0.440968 0.780102 0.637082 0.955229 0.101523 2003 1.093054 3,153 0.453853 0.739851 0.599173 0.913558 0.105743 2004 1.53297 3,104 0.465206 0.893362 0.726105 1.099147 0.103928 2005 1.59833 2,870 0.521603 0.995332 0.830641 1.192677 0.090625 2006 1.703548 2,720 0.531618 0.937315 0.779609 1.126921 0.092309 10

Figure 1. Coastal Logbook defined fishing areas with king mackerel regions indicated. 11

Figure 2. Revised king mackerel (1993-2006) nominal CPUE (solid circles), standardized CPUE (open diamonds) and upper and lower 95% confidence limits (dashed lines) of the standardized CPUE estimates for commercial vessels fishing hook and line gear (handline, electric reel, and trolling) in the Gulf of Mexico. 12

Figure 3. Revised king mackerel (1993-2006) nominal CPUE (solid circles), standardized CPUE (open diamonds) and upper and lower 95% confidence limits (dashed lines) of the standardized CPUE estimates for commercial vessels fishing hook and line gear in the Mixing Zone. 13

Figure 4. Revised king mackerel (1993-2006) nominal CPUE (solid circles), standardized CPUE (open diamonds) and upper and lower 95% confidence limits (dashed lines) of the standardized CPUE estimates for commercial vessels fishing hook and line gear in the south Atlantic. 14

Figure 5. Annual trend in the proportion of positive trips (A) and nominal CPUE (B) for the revised Gulf of Mexico 1993-2006 king mackerel commercial hook and line gear model. Data are plotted by season within years. A. B. Figure 6. Diagnostic plots for the binomial component of the revised Gulf of Mexico 1993-2006 king mackerel commercial hook and line gear model: A. the frequency distribution of the proportion positive trips; B. the Chi-Square residuals by year; C. the Chi-Square residuals by gear; and D. the Chi-Square residuals by hooks per line (effort1). A. B. C. D. 15

Figure 6 (continued). Diagnostic plots for the binomial component of the revised Gulf of Mexico 1993-2006 king mackerel commercial hook and line gear model: E. the Chi-Square residuals by area; F. the Chi-Square residuals by number of lines fished (numgear1); and G. the Chi-Square residuals by vessel length (ves_len). E. F. G. 16

Figure 7. Diagnostic plots for the lognormal component of the revised Gulf of Mexico 1993-2006 king mackerel commercial hook and line gear model: A) the frequency distribution of log(cpue) on positive trips, B) the cumulative normalized residuals (QQ-Plot) from the lognormal model. The red line is the expected normal distribution. A. B. Figure 8. Diagnostic plots for the lognormal component of the revised Gulf of Mexico 1993-2006 king mackerel commercial hook and line gear model: A. the Chi-Square residuals by year; B. the Chi-Square residuals by hooks per line (effort1); C. the Chi-Square residuals by area; and D. the Chi-Square residuals by hours fished (hrs_fished). A. B. C. D. 17

Figure 8 (continued). Diagnostic plots for the lognormal component of the revised Gulf of Mexico 1993-2006 king mackerel commercial hook and line gear model: E. the Chi-Square residuals by number of crew (crew1); F. the Chi-Square residuals by number of lines fished (Numgear1); G. the Chi-Square residuals by vessel length (ves_len); and H. the Chi-Square residuals by targeting (percent king mackerel in landings) E. F. G. H. 18

Figure 9. Annual trend in the proportion of positive trips (A) and nominal CPUE (B) for the revised Mixing Zone 1993-2006 king mackerel commercial hook and line gear model. A. B. Figure 10. Diagnostic plots for the binomial component of the revised Mixing Zone 1993-2006 king mackerel commercial hook and line gear model: A. the frequency distribution of the proportion positive trips; B. the Chi-Square residuals by year; C. the Chi-Square residuals by gear; and D. the Chi-Square residuals by area. A. B. C. D. 19

Figure 11. Diagnostic plots for the lognormal component of the revised Mixing Zone 1993-2006 king mackerel commercial hook and line gear model: A) the frequency distribution of log(cpue) on positive trips, B) the cumulative normalized residuals (QQ-Plot) from the lognormal model. The red line is the expected normal distribution. A. B. Figure 12. Diagnostic plots for the lognormal component of the revised Mixing Zone 1993-2006 king mackerel commercial hook and line gear model: A. the Chi-Square residuals by year; B. the Chi-Square residuals by hooks per line (effort1); C. the Chi-Square residuals by area; and D. the Chi-Square residuals by hours fished (hrs_fished). A. B. C. D. 20

Figure 12 (continued). Diagnostic plots for the lognormal component of the revised Mixing Zone 1993-2006 king mackerel commercial hook and line gear model: E. the Chi-Square residuals by targeting; F. the Chi- Square residuals by number of lines fished (numgear1); G. the Chi-Square residuals by vessel length (ves_len). E. F. G. 21

Figure 13. Annual trend in the proportion of positive trips (A) and nominal CPUE (B) for the revised South Atlantic 1993-2006 king mackerel commercial hook and line gear model. A. B. Figure 14. Diagnostic plots for the binomial component of the revised South Atlantic 1993-2006 king mackerel commercial hook and line gear model: A. the frequency distribution of the proportion positive trips; B. the Chi-Square residuals by year; C. the Chi-Square residuals by gear; and D. the Chi-Square residuals by area. A. B. C. D. 22

Figure 14 (continued). Diagnostic plots for the binomial component of the revised South Atlantic 1993-2006 king mackerel commercial hook and line gear model: E. the Chi-Square residuals by number of lines fished (numgear1) and F. the Chi-Square residuals by hooks per line (effort1). E. F. 23

Figure 15. Diagnostic plots for the lognormal component of the revised South Atlantic 1993-2006 king mackerel commercial hook and line gear model: A) the frequency distribution of log(cpue) on positive trips, B) the cumulative normalized residuals (QQ-Plot) from the lognormal model. The red line is the expected normal distribution. A. B. Figure 16. Diagnostic plots for the lognormal component of the revised South Atlantic 1993-2006 king mackerel commercial hook and line gear model: A. the Chi-Square residuals by year; B. the Chi-Square residuals by targeting (percent king mackerel in landings); C. the Chi-Square residuals by area; and D. the Chi-Square residuals by number of lines fished (numgear1). A. B. C. D. 24

Figure 16 (continued). Diagnostic plots for the lognormal component of the revised South Atlantic 1993-2006 king mackerel commercial hook and line gear model: E. the Chi-Square residuals by hooks per line (effort) and E. the Chi-Square residuals by hours fished (hrs_fished). E. F. 25